In Vivo Evolution of Policies that Govern a Distributed System

This paper addresses an important open problem confronting any decentralized and stateful access control (AC) mechanism for networked systems, particularly when the system at hand is large, heterogeneous and open. The problem, in a nutshell, is how to enable safe evolution of the policy that governs a given system, while that system continues to operate. This problem, and its solution, are addressed here in terms of the law-governed interaction (LGI) mechanism, but the ideas underlying this paper should be broadly relevant to other decentralized and stateful control mechanisms, such as the use of distributed firewalls for the protection of distributed enterprise systems.